LOURIE, Circuit Judge.
Finnigan Corporation appeals from the final determination of the United States International Trade Commission that claims 1-4, 8, 12, 14, and 17 of U.S. Patent 4,540,884 were not literally infringed and that claims 1-4 and 8 were anticipated under 35 U.S.C. § 102(b). See In re Certain Ion Trap Mass Spectrometers and Components Thereof, USITC Inv. No. 337-TA-393 (Dep’t Commerce Feb. 25, 1998) (“Initial Determination”); Certain Ion Trap Mass Spectrometers and Components Thereof; Notice of Final Commission Determination of No Violation of Section 337 of the Tariff Act of 1930, 62 Fed.Reg. 19946 (Dep’t Commerce Apr. 13, 1998) (“Final Determination”). Because the Commission did not err in its construction of the claims and its determination that those claims were not literally infringed, but did err in its conclusion that the claims were anticipated, we affirm-in-part and reverse-in-part.
[1357]*1357BACKGROUND
A. The Patented Technology
Finnigan is the assignee of the ’884 patent, which pertains to a method for using a “quadrupole ion trap” to generate a mass spectrum of a trapped sample for purposes of analysis. The. disclosed “quadrupole” ion trap is depicted in Figure 1 of the patent:
[[Image here]]
The ion trap generally works as follows: The sample to be analyzed is ionized to impart an electrical charge to the atoms or molecules of the sample. The ionized sample is then subjected to an electric field, called a “quadrupole field” or “trapping field,” which is generated by applying a radio-frequency (RF), alternating-current (AC) voltage to electrode ring 11. The magnitude of this AC voltage is denoted as “V” in the patent and its angular frequency is denoted as “<o” (the Greek letter “omega”). A direct-current (DC) voltage “U” may be added to the AC voltage. The quadrupole field causes certain ions to orbit within the storage region of the trap, which is bounded by the electrode ring and end caps 12 and 13. The extent of the orbit of a given ion depends upon its mass-to-charge (m/e) ratio.
The disclosed ion trap can be used as a mass spectrometer by changing or “scanning” the parameters that define the quadrupole field, viz., U, V or w, such that the trajectories of certain trapped ions are no longer sustainable within and leave the field where they are detected. The specification sets forth the process as follows:
DC and RF voltage (U, and V cos cot) are applied to a three-dimensional electrode structure such that ions over the entire specific mass range of interest are simultaneously trapped within the field imposed by the electrodes. Ions are then created or introduced into the quadrupole field area by any one of a variety of well known techniques. After this storage period, the DC voltage, U, the RF voltage[,] V, and the RF frequency,, are changed, either in combination or singly so that trapped ions of consecutive specific masses become successively unstable. As each trapped ionic species becomes unstable, all such ions develop trajectories that exceed the boundaries of the trapping field. These ions pass out of the trapping field through perforations in the field imposing electrode structure and impinge on a detector such as an electron multiplier or a Faraday collector. The detected ion current signal intensity as [a] function of time corresponds to a mass spectr[um] of the ions that were initially trapped.
’884 patent, col. 4, ll. 28-48.
The specification identifies the ion species that are trappable in a given quadrupole field by reference to a “Mathieu Stability Diagram,” an example of which is shown in Figure 4 of the patent:
[1358]*1358[[Image here]]
The physics giving rise to the bounded area of this diagram is complicated; fortunately a detailed understanding is unnecessary to resolve the controversy presented here. Simply stated, if a given ion in a given quadrupole field maps within the boundaries of the diagram, it is trappable; if a given ion maps outside the boundaries of the diagram, it is not trappable. In the words of the patent specification:
If [the] scanning parameters combine to map inside the stability envelope[,] then the given particle has a stable trajectory in the defined field. A charge[d] particle having a stable trajectory in a three[-]dimensional quadrupole field is constrained to an aperiodic orbit about the center of the field. Such particles can be thought of as trapped by the field. If for a particle m/e, U, V, r0 and 6 combine to map outside the stability envelop on the stability diagram, then the given particle has an unstable trajectory in the defined field. Particles having unstable trajectories in a three[-]dimensional field attain displacements from the center of the field which approach infinity over time. Such particles can be thought of as escaping the field and are consequently considered untrappable.
’884 patent, col. 3, l. 56 to col. 4, l. 2.
The foregoing aspects of the written description are reflected in independent claim 1, with the disputed claim language emphasized:
1. The method of mass analyzing a sample which comprises the steps of defining a three dimensional quadrupole field in which sample ions over the entire mass range of interest can be simultaneously trapped introducing or creating sample ions into the quadrupole field whereby ions within the range of interest are simultaneously trapped changing the three dimensional trapping field so that the simultaneously trapped ions of consecutive specific masses become sequentially unstable and leave the trapping field and detecting the sequential unstable ions as they leave the trapping field and providing an output signal indicative of the ion mass.
Independent claims 8 and 17 are similar to claim 1. The language in those claims corresponding to the disputed limitation in claim 1 reads as follows:
scanning one or more of U, V and between predetermined limits so that the trapped ions of specific mass become sequentially and selectively unstable and sequentially exit from the ion trap (claim 8)
sequentially selecting by ejecting ion[s] of different mass values by scanning the three dimensional quadrupole field (claim 17)
[1359]*1359B. The Accused Device and Technique
Bruker’s ESQUIRE-LC spectrometer operates in a manner similar to the disclosed embodiment of the ’884 patent in all material respects, except in the manner in which ions are ejected from the trapping field for detection. As previously mentioned, the patented embodiment ejects ions from the trap by changing or “scanning” the quadrupole field to render trapped ions unstable with respect thereto, ie., by transferring previously trapped ions outside of the stability diagram. In addition, the end caps 12 and 13 of the patented embodiment are grounded. See Figure 1 supra. The ESQUIRE device, in contrast, ejects ions without transferring the ions outside of the stability diagram. Instead, the trapped ions are ejected by applying a “supplementary” AC voltage to one of the end caps* This arrangement is depicted in Bruker’s “Primer” on the operation of the ESQUIRE machine, with the supplementary AC voltage labeled “auxiliary dipolar amplitude”:
Joint App. at A3300.
In the Bruker technique, certain ions are trapped in the storage area, just as in the patented embodiment. Each of these ions orbits within the trap with a “secular” frequency, which is dependent on the quadrupole scanning parameters. During analysis, one of these parameters (usually V) is changed to bring the secular frequency of a given ion into parity with the fixed frequency of the supplementary voltage (or a harmonic thereof). When this happens, the field produced by the supplementary voltage and the ion are in “resonance.” Resonance causes the ion to gain kinetic energy from the supplementary field, “just as a parent can impart additional energy to a child on a swing by pushing the child at the same moment on each ‘orbit’ of the swing. As the ion gains energy, its trajectory, like that of the swinging child, will increase over time, until the ion exits the trap ... and strikes a detector.” Commissioner’s Brief at 7. Eventually, as scanning continues, the secular frequency of another ion species is brought into parity with the fixed supplemental voltage frequency and is ejected, and so on. The important point is that, in contradistinction to the patented embodiment, the ions ejected from the trapping field are stable from the standpoint of the stability diagram.
Because the Bruker technique involves principles of resonance, it is referred to in the literature as “resonance ejection,” whereas the patented embodiment employs a form of “nonresonance ejection.”
C. The Proceedings Before the Commission
Finnigan filed a complaint at the Commission asserting that the Interven[1360]*1360ors1 were importing the ESQUIRE systems in violation of 19 U.S.C. § 1337(a) (1994) (“Section 337”) because these systems infringed, inter alia, its ’884 patent. See 19 U.S.C. § 1337(a)(1)(B) (prescribing that importation of articles that “infringe a valid and enforceable United States patent” is an unlawful trade practice). In response, the Commission instituted an investigation on February 20, 1997, see Certain Ion Trap Mass Spectrometers and Components Thereof; Notice of Investigation, 62 Fed.Reg. 8774 (1996), which culminated in a 309-page Initial Determination in which the Administrative Law Judge (ALJ) recommended that the Commission conclude that Section 337 had not been violated. As relevant here, the ALJ determined that claims 1-4, 8, 12, 14, and 17 were not literally infringed2 and that claims 1-4 and 8 were anticipated and therefore invalid under § 102(b). These determinations were adopted by the Commission.3
The ALJ’s noninfringement determination followed from his conclusion that the claims did not cover the resonance ejection technique practiced by the Bruker device. Much of the ALJ’s discussion of claim construction focused on the interpretation of the term “unstable,” which appears in independent claims 1 and 8. The ALJ found it “significant” that the specification makes “use of the terms ‘stable’ and ‘unstable’ with specific reference to the curves of the [Mathieu] stability diagram.... ” Initial Determination at 26. The ALJ also found significant the specification’s discussion of the prior art, which, like the Bruker device, involved “applying a voltage pulse between the end caps” to eject “trapped stable ions,” see ’884 patent, col. 2, ll. 1-10:
It is clear that for the purpose of the ‘884 patent the patentee made a distinction between the inventive mode of operation which relies on the ‘scanning parameters’ (U, V, co, and r0) and the prior art mode of operation which ejects ‘stable’ ions by applying a voltage pulse between the end caps.
Initial Determination at 25-26. The ALJ concluded that “the ’884 patent uses the terms ‘stable’ and ‘unstable’ to refer respectively[ ] to ions that are ‘trapped’ (stable) or untrappable (unstable), according to the theory set forth in the ... Mathieu stability diagram” to the exclusion of resonance ejection techniques that involve application of a supplementary voltage to the end caps to eject otherwise stable, trapped ions.4 See id. at 30, 45. Accordingly, the ALJ held claims 1-4, 8, 12, 14, and 17 not infringed. See id. at 162.
The ALJ found claims 1-4 and 8 to be anticipated for two reasons. First, he found that the subject matter of those claims was disclosed in an article written by Keith B. Jefferts. The ion trap used by Jefferts is depicted in Figure 2 of that article:
[1361]*1361[[Image here]]
Aside from the schematic disclosure of “Trap Drive Voltages” in Figure 2, the Jefferts article contains no disclosure concerning how ions are ejected from the trap — i.e., by nonresonance ejection (which the parties agree would anticipate the claims) or by resonance ejection (which would not anticipate). Although the article is thus unclear on this crucial point, the ALJ found that “the testimony developed at the hearing established that[,] especially to one skilled in the art, there is no doubt that mass selective instability [ie., nonre-sonance ejection], rather than resonant ejection, is the method disclosed in the Jefferts article.” Initial Determination at 67-68. In support of that finding, the ALJ relied on the testimony of Jefferts and on his conclusion that “there is no indication in the Jefferts article of an AC voltage supplied to the end caps for resonan[ce] ejection.” Id. at 68. He therefore concluded that the ejection technique must involve variation of the trap parameters as required by the claims in dispute.5
The second reason that the ALJ found the claims anticipated was Jefferts’ asserted prior public use of the claimed invention. The primary basis for the ALJ’s conclusion concerning prior use was Jef-ferts’ testimony. In response to Finni-gan’s argument that this testimony was insufficient absent other corroborating evidence to invalidate the claims, the ALJ noted that Jefferts’ credibility did “not appear to be open to any serious question” because he was not an interested party. See id. at 74. The ALJ also found many of the details of Jefferts’ work to be recorded in the 1968 article, which the ALJ characterized as not inconsistent with his testimony. See id. After separately finding Jefferts’ use to be sufficiently “public,” see id. at 75-78, the ALJ concluded that the evidence was clear and convincing that claims 1-4 and 8 were invalid, see id. at 78-79.
Finnigan appealed the Commission’s determination to this court. We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(6) (1994).
DISCUSSION
A. Standards of Review
We review factual findings of the Commission under the “substantial evidence” standard. See 19 U.S.C. § 1337(c) [1362]*1362(1994) (stating that those adversely affected by a Section 337 determination may appeal to this court for review in accordance with chapter 7 of Title 5); 5 U.S.C. § 706(2)(E) (1994) (setting forth the “substantial evidence” standard of review); Intel Corp. v. United States Int’l Trade Comm’n, 946 F.2d 821, 832, 20 U.S.P.Q.2d 1161, 1171 (Fed.Cir.1991). Under this standard, we will not disturb the Commission’s factual findings if they are supported by “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Surface Tech., Inc. v. United States Int’l Trade Comm’n, 801 F.2d 1336, 1340-41, 231 U.S.P.Q. 192, 196 (Fed.Cir.1986). We review the Commission’s legal determinations de novo. See 5 U.S.C. § 706(2)(A) (1994); YBM Magnex, Inc. v. International Trade Comm’n, 145 F.3d 1317, 1320, 46 U.S.P.Q.2d 1843, 1845 (Fed.Cir.1998).
“An infringement analysis entails two steps. The first step is determining the meaning and scope of the patent claims asserted to be infringed. The second step is comparing the properly construed claims to the device accused of infringing.” Markman v. Westview Instruments, Inc., 52 F.3d 967, 976, 34 U.S.P.Q.2d 1321, 1326 (Fed.Cir.1995) (en banc), aff’d, 517 U.S. 370, 116 S.Ct. 1384, 134 L.Ed.2d 577, 38 USPQ2d 1461 (1996). The first step, claim construction, is a question of law which we review de novo. See Cybor Corp. v. FAS Techs., Inc., 138 F.3d 1448, 1456, 46 U.S.P.Q.2d 1169, 1174 (Fed.Cir.1998) (en banc). The second step is factual. See North Am. Vaccine, Inc. v. American Cyanamid Co., 7 F.3d 1571, 1574, 28 U.S.P.Q.2d 1333, 1335 (Fed.Cir.1993).
Whether a prior art reference anticipates a patent claim is a question of fact. See In re Paulsen, 30 F.3d 1475, 1478, 31 USPQ2d 1671, 1674 (Fed.Cir.1994). Whether a claim limitation is inherent in a prior art reference for purposes of anticipation is also a question of fact. See In re Schreiber, 128 F.3d 1473, 1477, 44 U.S.P.Q.2d 1429, 1431 (Fed.Cir.1997).
B. Claim Construction & Literal Infringement of Independent Claim 17
Before addressing infringement and validity of independent claims 1 and 8, we first address and dismiss Finnigan’s argument concerning the construction of independent claim 17. Finnigan asserts that claim 17 does not contain the term “unstable” and that the ALJ therefore erred in construing this claim to be commensurate in scope with claims 1 and 8, which do contain this limitation. When claim 17 is properly construed, Finnigan continues, the ESQUIRE device infringes because it “sequentially select[s] by ejecting ion[s] of different mass values by scanning the three dimensional quadrapole field” as that claim requires. The Commission responds that Finnigan’s argument concerning the construction of claim 17 cannot be heard by this court because it was not raised in Finnigan’s petition to the Commission. Finnigan rejoins that this court may consider its claim construction argument because, under Cybor Corp. v. FAS Technologies, Inc., 138 F.3d 1448, 46 U.S.P.Q.2d 1169 (Fed.Cir.1998) (en banc), claim construction is a question of law that receives de novo review on appeal.
We agree with the Commission that the argument concerning claim 17 has been waived. A party seeking review in this court of a determination by the Commission must “specifically assert” the error made by the ALJ in its petition for review to the Commission. See Personalized Media Communications, L.L.C. v. International Trade Comm’n, 161 F.3d 696, 709, 48 U.S.P.Q.2d 1880, 1891 (Fed.Cir.1998); Checkpoint Sys., Inc. v. United States Int’l Trade Comm’n, 54 F.3d 756, 760, 35 U.S.P.Q.2d 1042, 1046 (Fed.Cir.1995). This rule is required by principles of administrative law and by the Code of Federal Regulations:
[Ojrderly procedure and good administration require that objections to the proceedings of an administrative agency be made while it has opportunity for correction in order to raise issues re[1363]*1363viewable by the courts. When a party to an administrative proceeding fails to assert an issue during that proceeding, other parties to the proceeding are not given notice and an opportunity to present arguments before the agency; moreover, the agency is not afforded the opportunity to consider the issue in the first instance. Thus, simple fairness to those who are engaged in the tasks of administration, and to litigants, requires as a general rule that courts should not topple over administrative decisions unless the administrative body not only has erred but has erred against objection made at the time appropriate under its practice.
Checkpoint, 54 F.3d at 760, 35 U.S.P.Q.2d at 1045 (citations, quotations, and brackets omitted); see also id. (noting that 19 C.F.R. § 210.54(a) (1994), now 19 C.F.R. § 210.43(b) (1998), requires that a petition for review ’contain a “concise argument providing the reasons why review by the Commission is necessary,” and specifying that “[a]ny issue not raised in the petition for review-will be deemed to have been abandoned.”).
Finnigan’s petition for review did not assert that claims 1, 8, and 17 were of different scope in any relevant manner, and did not argue the relevance of the lack of the term “unstable” in claim 17. See Joint App. at A5891-907. Therefore, Finnigan’s argument concerning the construction of claim 17 has been waived and we will not consider it. See Checkpoint, 54 F.3d at 760, 35 U.S.P.Q.2d at 1046 (holding that this court would not address a party’s claim construction argument when it had not included it in its petition for review).
Cybor does not help Finnigan. That this court may review a lower tribunal’s claim construction de novo under Cybor does not require us to consider claim construction arguments that were not raised before the Commission. A party’s argument should not be a moving target. The argument at the trial and appellate level should be consistent, thereby ensuring a clear presentation of the issue to be resolved, an adequate opportunity for response and evidentiary development by the opposing party, and a record reviewable by the appellate court that is properly crystallized around and responsive to the asserted argument. See Sage Prods., Inc. v. Devon Indus., Inc., 126 F.3d 1420, 1426, 44 U.S.P.Q.2d 1103, 1108 (Fed.Cir.1997). Cybor does not require a different result. While Cybor confirmed this court’s ability to review claim construction de novo, it does not require us to effectively retry claim construction de novo by consideration of novel arguments not first presented to the tribunal whose decision is on review.
Because Finnigan did not present to the Commission the argument concerning the scope of claim 17 that it currently presses on appeal, we decline to consider this argument.
C. Claim Construction & Literal Infringement of Independent Claims 1 and 8
Finnigan argues that the ALJ erred in construing the term “unstable” in claims 1 and 8 with reference to the stability diagram disclosed in the ’884 patent. Finnigan asserts that the specification makes clear that an “unstable” ion is any ion that escapes the quadrupole field, i.e., any ion whose trajectory exceeds the trapping space. Finnigan also asserts that it was improper for the ALJ to limit the claims to the disclosed embodiment. The Commission responds that the specification is clear that the term “unstable” refers to an ion that maps outside of the stability diagram. The Commission also asserts that, because the specification does not disclose the resonance ejection technique, the claims may not be interpreted to encompass it without running afoul of the written description and enablement requirements of 35 U.S.C. § 112, ¶ 1. Both parties also argue that the prosecution history and evidence extrinsic to the patent support their respective positions.
When construing a claim, a court principally consults the evidence intrinsic [1364]*1364to the patent, viz., the claims themselves, the written description portion of the specification, and the prosecution history. See Vitronics Corp. v. Conceptronic, Inc., 90 F.3d 1576, 1582-83, 39 U.S.P.Q.2d 1573, 1576-77 (Fed.Cir.1996). Reference to this evidence, particularly the specification, is more than sufficient to resolve the claim construction issue presented here. The specification explicitly defines “stable” ions as those that map within the Mathieu stability diagram and are therefore trapped by the quadrupole field, and “unstable” ions as those that do not. We have already quoted the relevant portion of the specification above:
If [the] scanning parameters combine to map inside the stability envelope!,] then the given particle has a stable trajectory in the defined field .... Such particles can be thought of as trapped by the field. If for a particle m/e, U, V, r0 and combine to map outside the stability envelop on the stability diagram, then the given particle has an unstable trajectory in the defined field .... Such particles can be thought of as escaping the field and are consequently considered untrappable.
’884 patent, col. 3, l. 56 to col. 4, l. 2 (emphasis added).
A patentee may be his or her own lexicographer. See, e.g., Vitronics, 90 F.3d at 1582, 39 U.S.P.Q.2d at 1576. The inventors therefore could have defined an “unstable” ion in the specification as one that is ejected from the quadrupole field, but they did not do so. In fact, the discussion of the prior art in the specification, which is similar in operation to the Bruker device inasmuch as it involves ejection of ions via the application of voltages to the end caps of the trap, belies Finnigan’s contention that all ejected ions are necessarily “unstable” ions. See ’884 patent, col. 2., ll. 5-8 (describing the prior art: by “applying a voltage pulse between the end caps, the trapped stable ions are ejected out of the storage region to a detector.”) (emphasis added).
Thus, we agree with the Commission that the specification requires the ALJ’s claim construction — i.e., that the claim term “unstable” refers to ions that are unstable from the perspective of the disclosed Mathieu stability diagram.6 We have considered the other evidence cited to us by the parties — specifically other portions of the specification, the prosecution history, and certain extrinsic evidence— but find such evidence vague and unhelpful as to the meaning of the term “unstable.” Because Finnigan does not contest that there is no infringement if the ALJ’s claim construction is adopted by this court, we affirm the ALJ’s determination of nonin-fringement of independent claims 1 and 8 and all claims that are dependent thereupon. See Wahpeton Canvas Co., Inc. v. Frontier, Inc., 870 F.2d 1546, 1553, 10 U.S.P.Q.2d 1201, 1208 (Fed.Cir.1989) (“It is axiomatic that dependent claims cannot be found infringed unless the claims from which they depend have been found to have been infringed.”).
D. Anticipation
Finnigan argues that neither of the ALJ’s grounds for finding claims 1-4 and 8 anticipated is sustainable. As to the Jef-ferts article, Finnigan asserts that the article does not clearly disclose nonresonance ejection. Finnigan points to Jefferts’ testimony in support of its argument, testimony which Finnigan characterizes as ambiguous concerning the reference’s disclosure. As to Jefferts’ public use, Finnigan argues that Jefferts’ uncorroborated testimony cannot constitute clear and convincing evidence of invalidity as a matter of law. The Commission responds that substantial evidence supports the ALJ’s finding that the Jefferts article, as supplemented by Jef-ferts’ testimony, anticipates the disputed claims. The Commission asserts further [1365]*1365that Jefferts was an uninterested party in the present dispute, and therefore that his testimony concerning his public use of the claimed invention need not be corroborated. Moreover, the Commission continues, Jefferts’ article corroborated his testimony.
Patents are presumed valid by statute. See 35 U.S.C. § 282 (1994). “The burden is on the party asserting invalidity to prove it with facts supported by clear and convincing evidence.” SSIH Equip., S.A v. United States Int’l Trade Comm’n, 718 F.2d 365, 375, 218 U.S.P.Q. 678, 687 (Fed.Cir.1983). We have reviewed the record and have determined that the invalidity determination was not based upon substantial evidence which is clear and convincing that the Jefferts article or Jefferts’ public use anticipates. We therefore reverse the Commission’s determination that claims 1-4 and 8 of the ’884 patent are invalid.7
1. The Jefferts Article
A prior art reference anticipates a patent claim if the reference discloses, either expressly or inherently, all of the limitations of the claim. See Kalman v. Kimberly-Clark Corp., 713 F.2d 760, 771, 218 U.S.P.Q. 781, 789 (Fed.Cir.1983)., The obvious shortcoming of the Jefferts article is that it does not expressly disclose the use of nonresonance ejection, a limitation of claims 1 and 8 as we have construed them. The ALJ was able to close this gap between the Jefferts article and the claim by concluding that “one skilled in the art” would understand the article to disclose nonresonance ejection. Because of the ALJ’s reliance on the understanding of one skilled in the art, we understand the ALJ to have relied on an inherency theory in finding the claims anticipated.
The operation of inherency in anticipation was aptly explained by this court in Continental Can Co., U.S.A. v. Monsanto Co., 948 F.2d 1264, 20 U.S.P.Q.2d 1746 (Fed.Cir.1991):
To serve as an anticipation when the reference is silent about the asserted inherent characteristic, such gap in the reference may be filled with recourse to extrinsic evidence. Such evidence must make clear that the missing descriptive matter is necessarily present in the thing described in the reference, and that it would be so recognized by persons of ordinary skill. In re Oelrich, 666 F.2d 578, 581, 212 U.S.P.Q. 323, 326 (CCPA 1981) (quoting Hansgirg v. Kemmer, 26 C.C.P.A. 937, 102 F.2d 212, 214, 40 U.S.P.Q. 665, 667 (1939)) [states]:
Inherency, however, may not be established by probabilities or possibilities. The mere fact that a certain thing may result from a given set of circumstances is not sufficient. If, however, the disclosure is sufficient to show that the natural result flowing from the operation as taught would result in the performance of the questioned function, it seems to be well settled that the disclosure should be regarded as sufficient.
This modest flexibility in the rule that “anticipation” requires that every element of the claims appear in a single reference accommodates situations where the common knowledge of technologists is not recorded in the reference; that is, where technological facts are known to those in the field of the invention, albeit not known to judges.
Continental Can, 948 F.2d at 1268-69, 20 U.S.P.Q.2d at 1749-50. The ALJ’s understanding of the knowledge of one skilled in the art was informed solely by the testimony of Jefferts. The relevant portions of that testimony are as follows:
Jefferts: “[Figure 2] is a diagram that’s appropriate to the first experiment that I did in Bell Laboratories.”
ALJ: “All right. Now, were you using a nonresonant ejection technique in your first experiment ... [a]t Bell Laboratories?”
[1366]*1366Jefferts: “At least part of the time, yes.”
ALJ: “And does [Figure 2] express that or depict that experiment?”
Jefferts: “I think so.”
ALJ: “Could you explain that?”
Jefferts: “I think it can be interpreted to mean nonresonant ejection almost exclusively. It shows no orbit drive, as it were, or associated RF field to drive the ions out. On the other hand, this is a pretty general box, and if it — ”
ALJ: “Which is the general box?”
Jefferts: “That says ‘trap drive voltage.’ And the fact is that it’s the same diagram also applied to later versions of this system, wherein I did selective excitation and selective ejection [i.e., resonance ejection].”
Joint App. at A2034 (emphasis added). The portions of this testimony that we have emphasized show that Jefferts’ testimony was far from unequivocal. In the end, Jefferts as much as admitted that Figure 2 might disclose a set-up for performing either resonance or nonresonance ejection. The mere possibility that Figure 2 might be understood by one of skill in the art to disclose nonresonance ejection is insufficient to show that it is inherently disclosed therein. See Continental Can, supra. As such, because one skilled in the art would not necessarily recognize that nonresonance ejection is disclosed in the Jefferts article, the evidence is not clear and convincing that the Jefferts article inherently anticipates claims 1 and 8.
2. Jefferts’Alleged Public Use
The law has long looked with disfavor upon invalidating patents on the basis of mere testimonial evidence absent other evidence that corroborates that testimony. The Supreme Court recognized over one hundred years ago that testimony concerning invalidating activities can be “unsatisfactory” due to “the forgetfulness of witnesses, their liability to mistakes, their proneness to recollect things as the party calling them would have them recollect them, aside from the temptation to actual perjury.” The Barbed-Wire Patent, 143 U.S. 275, 284, 12 S.Ct. 443, 36 L.Ed. 154 (1892). Accordingly, “[witnesses whose memories are prodded by the eagerness of interested parties to elicit testimony favorable to themselves are not usually to be depended upon for accurate information,” and therefore such testimony rarely satisfies the burden upon the interested party, usually the accused infringer, to prove invalidity by clear and convincing evidence. See id.8
Mere testimony concerning invalidating activities is received with further skepticism because such activities are normally documented by tangible evidence such as devices, schematics, or other materials that typically accompany the inventive process. See Woodland Trust v. Flowertree Nursery, Inc., 148 F.3d 1368, 1373, 47 U.S.P.Q.2d 1363, 1367 (Fed.Cir.1998) (noting that the skepticism with which mere testimony of invalidating activity is received is “reinforced, in modern times, by the ubiquitous paper trail of virtually all [1367]*1367commercial activity. It is rare indeed that some physical record {e.g., a written document such as notes, letters, invoices, notebooks, or a sketch or drawing or photograph showing the device, a model, or some other contemporaneous record) does not exist.”); Eibel Process Co. v. Minnesota & Ontario Paper Co., 261 U.S. 45, 60, 43 S.Ct. 322, 67 L.Ed. 523 (1923) (holding that the oral testimony of prior public use “falls short of being enough to overcome the presumption of novelty from the granting of the patent” when “there is not a single written record, letter or specification of prior date to [the patentee’s] application that discloses any such discovery by anyone. ...”).
While this court has in the past applied the requirement of corroboration more often in the context of priority disputes under 35 U.S.C. § 102(g),9 corroboration has been required to prove invalidity under other subsections of § 102 as well.10 In the context of § 102(f) (derivation) and § 102(g) (priority), we have stated that “the case law is unequivocal that an inventor’s testimony respecting facts surrounding a claim of derivation or priority of invention cannot, standing alone, rise to the level of clear and convincing proof.” Price v. Symsek, 988 F.2d 1187, 1194, 26 U.S.P.Q.2d 1031, 1036 (Fed.Cir.1993). No principled reason appears for applying a different rule when other subsections of § 102 are implicated: a witness’s uncorroborated testimony is equally suspect as clear and convincing evidence if he testifies concerning the use of the invention in public before invention by the patentee (§ 102(a)), use of the invention in public one year before the patentee filed his patent (§ 102(b)), or invention before the pat-entee (§ 102(g)).
Moreover, the need for corroboration exists regardless whether the party testifying concerning the invalidating activity is interested in the outcome of the litigation {e.g., because that party is the accused infringer) or is uninterested but testifying on behalf of an interested party. That corroboration is required in the former circumstance cannot be debated. See, e.g., Stevenson v. International Trade Comm’n, 67 C.C.P.A. 109, 612 F.2d 546, 550, 204 U.S.P.Q. 276, 280 (1979) (“Uncorroborated oral testimony of prior inventors or users with a demonstrated financial interest in the outcome of the litigation is insufficient to provide such proof.”). Uninterested witnesses are also subject to the corroboration requirement. For example, in Barbed-Wire Patent, some twenty-four [1368]*1368witnesses, all apparently uninterested in the outcome of the case, testified on behalf of the accused infringer that they had seen the patented fence exhibited by a third party, Mr. Morley, at a county fair more than two years prior to the filing of the patent. See Barbed-Wire Patent, 143 U.S. at 286-87, 12 S.Ct. 443. That the witnesses themselves were not interested did not immunize their testimony from the corroboration requirement. See Barbed-Wire, 143 U.S. at 284, 12 S.Ct. 443 (“Witnesses whose memories are prodded by the eagerness of interested parties to elicit testimony favorable to themselves are not usually to be depended upon for accurate information.”) (emphasis added). It is not surprising that the cases have held that testimony concerning a witness’s own anticipatory activities must be corroborated. A witness who testifies to antedating the invention of the patent-in-suit can be expected to derive a sense of professional or personnel accomplishment in being the first in the field, and in this sense is not uninterested in the outcome of the litigation, even if that witness is not claiming entitlement to a patent. Of course, the need for corroboration takes on special force when an otherwise uninterested witness shows some reason to be biased in favor of the interested party:
As we have had occasion before to observe, oral testimony, unsupported by patents or exhibits, tending to show pri- or use of a device regularly patented, is, in the nature of the case, open to grave suspicion. The Barbed Wire-Patent, 143 U.S. 275, 12 S.Ct. 443, 36 L.Ed. 154 [ (1892) ]. Granting the witnesses to be of the highest character, and never so conscientious in their desire to tell only the truth, the possibility of their being mistaken as to the exact device used, which, though bearing a general resemblance to the one patented, may differ from it in the very particular which makes it patentable, is such as to render oral testimony peculiarly untrustworthy; particularly so if the testimony be taken after the lapse of years from the time the alleged anticipating device was used. If there be added to this a personal bias, or an incentive to color the testimony in the interest of the paody calling the witness, to say nothing of downright perjury, its value is, of course, still more seriously impaired.
Deering v. Winona Harvester Works, 155 U.S. 286, 300-01, 15 S.Ct. 118, 39 L.Ed. 153 (1894) (emphasis added). In the final analysis, the Supreme Court has defined the necessity of corroboration not with reference to the level of interest of the testifying witness, but rather because of doubt that testimonial evidence alone in the special context of proving patent invalidity can meet the clear and convincing evidentiary standard to invalidate a patent.
Thomson S.A v. Quixote Corp., 166 F.3d 1172, 49 U.S.P.Q.2d 1530 (Fed.Cir.1999), cert. denied, — U.S. —, 119 S.Ct. 2395, — L.Ed.2d —, 67 U.S.L.W. 3671 (1999) , is not to the contrary. The Thomson court did opine on the necessity of corroboration, stating that “corroboration is required only when the testifying inventor is asserting a claim of derivation or priority of his or her invention and is a named party, an employee of or assignor to a named party, or otherwise is in a position where he or she stands to directly and substantially gain by his or her invention being found to have priority over the patent claims at issue.” Thomson, 166 F.3d at 1176, 49 U.S.P.Q.2d at 1533. However, Thomson did not involve uncorroborated testimony of a single witness. Indeed, the district court in that case “noted that the evidence supporting the anticipation finding came from one or more sources: the live testimony of two people who had worked on [the project that was alleged to anticipate]; an expert’s report and portions of his deposition testimony ...; the expert’s exhibits; and certain-documents that the expert had reviewed.” See id. at 1174, 166 F.3d 1172, 49 U.S.P.Q.2d at 1531. Therefore, the facts of Thomson did not present the question of the necessity of corroboration vel non, but rather the sufficiency of the corrobo[1369]*1369rating evidence, a distinct inquiry involving an assessment of the totality of the circumstances, including consideration of “the interest of the corroborating witness in the subject matter of the suit.” See Woodland Trust, 148 F.3d at 1371, 47 U.S.P.Q.2d at 1366 (citing In re Reuter, 670 F.2d 1015, 1021 n. 9, 210 U.S.P.Q. 249, 255 n. 9 (CCPA 1981)) 11; see also id. at 1373, 47 U.S.P.Q.2d at 1368 (“The relationship of the witnesses and the fact that the asserted prior uses ended twenty years before the trial, and were abandoned until the defendant reportedly learned of the patentee’s practices, underscore the failure of this oral evidence to provide clear and convincing evidence of prior knowledge and use.”). Cases like Thomson and Woodland Trust correctly recognized that the level of interest of the testifying witness is an important consideration when such testimony is offered to corroborate another witness’s testimony. Those cases, however, do not stand for the proposition that only an interested witness’s testimony requires corroboration. In any event, corroboration is required of any witness whose testimony alone is asserted to invalidate a patent, regardless of his or her level of interest. Cf. Price, 988 F.2d at 1194 & 1195 n. 3, 26 U.S.P.Q.2d at 1036 & 1037 n. 3.
Returning to the facts of this case, Jefferts’ testimony that he used the claimed invention more than one year prior to the filing of the ’884 patent was not corroborated by other evidence. The Jefferts’ article simply does not corroborate his testimony because, as we have noted, that article is ambiguous at best concerning the claimed use of nonresonance ejection. Similarly, other testimony taken before the Commission was relevant only to whether Jefferts’ experiments were sufficiently public to constitute public use. See Initial Determination at 77-78 (discussing the testimony of Drs. Ensberg and Klemperer). This testimony is therefore not corroborative of Jefferts’ use of the claimed invention.
The Commission cites certain statements in the case law in support of the proposition that not every aspect of Jefferts’ testimony needs to be corroborated. See Knorr v. Pearson, 671 F.2d 1368, 1374, 213 U.S.P.Q.2d 196, 201 (CCPA 1982) (“The law does not impose an impossible standard of ‘independence’ on corroborative evidence by requiring that every point [necessary to prove invalidity] be corroborated by evidence having a source totally independent of the inventor; indeed such a standard is the antithesis of the rule of reason.”); Kridl v. McCormick, 105 F.3d 1446, 1451, 41 U.S.P.Q.2d 1686, 1688 (Fed.Cir.1997) (“An evaluation of all pertinent evidence must be made so that a sound determination of the credibility of the inventor’s story may be reached.”). These statements, however, are inapt here because they pertained to situations in which the question to be resolved was whether, under the totality of the circumstances, certain evidence was sufficiently corroborative of otherwise uncorroborated testh mony of invalidating activities. See Knorr, 671 F.2d at 1374, 213 U.S.P.Q. at 201 (“[W]e hold that [Pearson’s reduction to practice] was adequately corroborated by independent evidence.”); Kridl, 105 F.3d at 1450, 41 U.S.P.Q.2d at 1689 (assessing whether Swain’s testimony was sufficiently corroborative of McCormick’s claim of priority). In this case, the sole basis to support a determination of a prior public use was Jefferts’ testimony concerning his own work; there was no evidence corroborative [1370]*1370of this testimony at all. This appeal therefore presents a different question from the “sufficiency” question presented in Knorr and Kridl. See note 11, supra, and accompanying text.
In the end, what we are left with is Jefferts’ testimony concerning his alleged public use. Such evidence is insufficient as a matter of law to establish invalidity of the patent. This is not a judgment that Jefferts testimony is incredible, but simply that such testimony alone cannot surmount the hurdle that the clear and convincing evidence standard imposes in proving patent invalidity. See Mahurkar v. C.R. Bard, Inc., 79 F.3d 1572, 1577, 38 U.S.P.Q.2d 1288, 1291 (Fed.Cir.1996) (noting that the corroboration rule, “provides a bright line for both district courts and the PTO to follow....”).
CONCLUSION
The Commission did not err in its claim construction and its subsequent determination that the accused devices do not literally infringe claims 1-4, 8, 12, 14, and 17 of the ’884 patent. The Commission did, however, err in concluding that claims 1-4 and 8 were anticipated by the Jefferts article and Jefferts’ alleged prior public use under § 102(b). Neither of these grounds for invalidity is supported by substantial evidence that is clear and convincing. The decision of the Commission is therefore
AFFIRMED-IN-PART and REVERSED-IN-PART.